Choosing an LLM provider

Every agent needs one LLM provider. You configure providers at Settings → Integrations → LLM Integrations → Add integration, where each provider has its own tab.

Anthropic

Needs an API key starting sk-ant-. Bringing your own Anthropic key is useful when you want cost reporting, higher rate limits, or billing kept separate. Gives you Claude Opus 4.7, Claude Sonnet 4.6 and Claude Haiku 4.5.

Codex (OpenAI)

Needs an OpenAI API key and bills per token. Gives you GPT 5.5, 5.4, 5.4 Mini, 5.3 Codex and 5.2.

OpenCode

Needs a Zen API key and acts as a gateway to a much wider set: Claude variants, GPT models, Gemini, GLM, Qwen, Kimi, MiniMax and more. Reach for this when you want to compare models without opening an account with each provider.

Choosing per agent

Each agent picks its own provider and model in the agent editor, so a cheap model can handle your digests while a stronger one reviews code. A change takes effect on the next task. Work already in progress finishes on the model it started with.